Structure of H2/O2/N2 Flames at Atmospheric Pressure Studied by ...

Structure of H2/O2/N2 Flames at Atmospheric Pressure Studied by Molecular Beam Mass
Spectrometry and Modeling
D.A. Knyazkov 1∗ , O.P. Korobeinichev 1,2 , A.G. Shmakov 1 , I.V. Rybitskaya 1,2 , T.A. Bolshova 1 ,
A.A. Chernov 1 , A.A.Konnov 3
1 Institute of Chemical Kinetics and Combustion, Novosibirsk, Russia
2 Novosibirsk State University, Novosibirsk, Russia
3 Vrije Universiteit Brussel, Brussels, Belgium
Abstract
Structure of laminar premixed flat H2/O2/N2 flames with different equivalence ratios at atmospheric pressure is
investigated experimentally and by numerical modeling. Concentration profiles of stable species (H2, O2, H2O) as
well as of H atoms and OH radicals in the flames were measured using molecular beam mass-spectrometry with soft
ionization by electron impact. A good agreement between the obtained experimental data and the results of
numerical simulation using three different detailed kinetic mechanisms indicates the robustness of these models in
predicting the structure of hydrogen-oxygen flames at atmospheric pressure.
Introduction
Through many years and up to now, mechanism and
chemistry of combustion of hydrogen-oxygen mixtures
have been the subject of many investigations [1-10, 27].
Chemical kinetic mechanism of combustion of H2/O2
mixtures is of great importance because it is an integral
part of kinetic models for hydrocarbons oxidation. The
progress in development of these mechanisms is based
on a comparison of experimental data (on burning
velocity, ignition delay, flame structure etc.) obtained
over wide range of conditions (pressure, temperature,
composition of fresh mixture) with the results of
numerical simulation performed using the testable
kinetic mechanism. In kinetic mechanisms, the reactions
with active species, atoms and radicals, play a key role.
Therefore, the most important, while a challenging task
is to measure their concentration and concentration
profiles in the flames. Probing molecular beam massspectrometry
with soft electron-impact ionization is one
of the most appropriate tools for this purpose. Eltenton
[8] as well as Foner and Hudson [9] were the pioneers,
who applied this technique for detection of atoms and
radicals in low-pressure flames. Probing molecular
beam mass-spectrometry (pMBMS) is the most useful
technique allowing all species in the flame (including
atoms and radicals) to be measured simultaneously. It
was widely used for studying the structure of flames,
including the flames of hydrogen-oxygen mixtures.
The structure of such flames was studied mostly at
subatmospheric pressures [10-13]. Only a few works
[14-16] were focused on studying the structure of
hydrogen-oxygen flames at atmospheric and higher
pressures, due to the difficulties of measurements of
species concentration profiles by pMBMS at these
conditions. The concentration profiles of stable species
only were measured in fuel-rich [14-15] and
stoichiometric [16] flames. To validate thoroughly a
chemical kinetic mechanism for combustion of
hydrogen at 1 atm, one needs the data on spatial
∗ Corresponding author: knyazkov@kinetics.nsc.ru
Proceedings of the European Combustion Meeting 2009
variation of concentration of active species (particularly,
H and OH) in the atmospheric hydrogen flames.
Specific Objectives
This work aims to investigate the structure of
H2/O2/N2 flames over the range of equivalence ratios at
atmospheric pressure by molecular beam mass
spectrometry and by modeling for validation of the
kinetic mechanisms developed earlier. One of the key
emphases of this study is to validate the pMBMSsystem
with soft electron-impact ionization for
determining the structure of hydrogen flames, in
particular, for measurement of concentration profiles of
the key flame species (H and OH).
Experimental method
The flames studied were stabilized under nearadiabatic
conditions on a perforated plate burner using the
heat flux method [17, 18]. The burner consists of a burner
head mounted on 35-cm-high tube with water jacket.
The burner head represents 3-mm-thick copper plate 24
mm in diameter perforated with the holes 0.5 mm in
diameter (centers of the holes are equally spaced 0.7
mm apart) and heating jacket. The heating jacket of the
burner head is supplied with thermostated water to
maintain the temperature of the burner plate constant at
333 K. The water jacket of the burner tube is supplied
with circulating water at 308 K, which controls the
temperature of a fresh gas mixture. The idea of the
heat flux method is based on finding of the flow
velocity of the fresh mixture through the burner when
the heat losses into the burner are compensated by the
controlled heating of the fresh mixture, and the balance
of the heat losses to the burner surface is close to zero.
The balance of the heat losses at the burner surface is
determined by measuring the radial temperature
distribution on the burner plate. For this purpose, six
copper-constantan thermocouples were soldered into the
burner holes at 0, 2.4, 4.5, 7, 10 and 12 mm from the

Table 1. Summary of different parameters for the flames studied.
Flame
Rich flame Near-stoichiometric
flame
Lean flame Calibration flame
Equivalence ratio
2 1.1 0.47 1.1
φ=([H2]/[O2])/([H2]/[O2])stoichiometry
Dilution ratio D=[O2]/([O2]+[N2]) 0.077 0.09 0.209 0.14
Flow velocity of fresh mixture at burner
surface (in the experiments), cm/s
40.7 42.5 34.24 49.1
Measured burning velocity of free
propagating flame, cm/s
41.1 47 55
Burning velocity of free propagating flame
calculated using kinetic mechanism [2], cm/s
39.6 38.3 36.2 141.6
center of the burner. The flow velocity of the fresh
mixture whereby the above conditions are satisfied is
adiabatic burning velocity by definition.
Compositions and flow rates of the fresh mixtures
studied were set by mass flow controllers (MKS
Instruments Inc.). The stated purity of gases (H2, O2, N2)
was 99.95%. Three flames of different compositions
were studied in this work and a flame called “calibration
flame” was used to calibrate the MBMS-setup for
measurements of absolute concentrations of H and OH
radicals. Table 1 shows the compositions, equivalence
and dilution ratios for all four flames. Linear velocities
of the fresh mixtures at the burner surface as well as
measured and calculated (using PREMIX code [19-21]
and chemical kinetic mechanism [2]) burning velocities
of these flames are also given in the Table 1. To
stabilize reliably the flames during the flame structure
measurements, the flow velocities of the fresh mixtures
were set somewhat below the burning velocities of these
mixtures. The lean flame (φ=0.47) had a cellular
structure when the flow of the fresh gas mixture is close
to the burning velocity. Thus, to eliminate this effect,
the flame was stabilized with the fresh mixture flow
40% lower than the burning velocity of the flame in this
mixture. Temperature profiles in the flames were
measured using Pt/Pt+10%Rh thermocouple made of
wires 0.02 mm in diameter and coated with SiO2 (total
diameter of the coated thermocouple was ~0.04 mm).
Concentration profiles of the species were obtained
using molecular beam sampling system coupled with
quadrupole mass-spectrometer MS-7302 with soft
ionization by electron impact [22]. Sampling was
performed at different heights above the burner surface
using a quartz “sonic” probe. The opening angle of the
probe was 40 ○ , the orifice diameter was 0.08 mm, and
the wall thickness near the orifice was 0.08 mm.
A sampling probe is known to perturb a flame
structure. The total probe effect on the flame is a
combination of thermal and gas-dynamic perturbations
of the flame. To take into account the thermal
perturbations, temperature profiles measured in the
flame by a thin thermocouple at a certain distance
“probe tip - thermocouple” in front of the probe tip are
usually used in the calculations of concentration profiles
of flame species. In our experiments, the distance
“probe tip – thermocouple junction” was chosen to be
0.2 mm. To validate this choice we have preliminary
measured temperature profiles in the nearstoichiometric
flame with the thermocouple positioned
2
at different distances from the probe tip (0.1, 0.2, 0.3
and 0.4 mm). The obtained results are shown in Fig. 1.
Using these profiles and the detailed chemical kinetic
mechanism [23] we simulated the flame structure. A
criterion for choosing the most appropriate distance
“probe tip – thermocouple junction” was as following:
position of H and OH maximums on corresponding
measured concentration profiles should be in agreement
with those on calculated profiles obtained using the
appropriately measured temperature profile. We find
this criterion to be justified, because in this case the
accuracy of absolute concentration of the species does
not need to be perfect. In addition, position of the
maximum on the measured H and/or OH concentration
profile is far easier to recognize and to compare with the
numerical results. Figure 2 shows H and OH
concentration profiles measured in the nearstoichiometric
flame as well as calculated ones using the
temperature profiles given in Fig. 1 (the calculations
were performed using PREMIX code [19-21] and
chemical kinetic mechanism [23]). Comparing the
experimental and modeling results shown in Fig. 2 one
can conclude that for the near-stoichiometric flame the
appropriate distance “probe tip – thermocouple
junction” is somewhere between 0.15 and 0.2 mm.
Since the conditions (post-flame temperature, velocity
of the flow riding on the probe) for lean (φ=0.47) and
rich (φ=2.0) H2/O2/N2 flames studied are very similar to
those for the near-stoichiometric flame (φ=1.1), the
Temperature, K
1600
1400
1200
1000
800
600
0.4 mm
0.3 mm
0.2 mm
0.1 mm
400
0.0 0.5 1.0 1.5 2.0
Distance from burner surface, mm
Fig. 1. Temperature profiles measured in presence of
the sampling probe by the thermocouple positioned
at different distances “probe tip – thermocouple
junction” in near-stoichiometric H2/O2/N2 flame
(φ=1.1, D=0.09).

Mole fraction of H
0.010
0.008
0.006
0.004
0.002
measured H profile
0.1 mm
0.2 mm
0.3 mm
0.4 mm
0.000
0 1 2 3 4 5
Distance from burner surfce, mm
Mole fraction of OH
3
0.0020
0.0015
0.0010
0.0005
measured OH profile
0.0000
0 1 2 3 4 5
Distance from burner surface, mm
appropriate position of the thermocouple relative to
probe tip can be assumed to be the same.
To take into account the gas-dynamic perturbation of
the flames by the sampling probe, the upstream shift ∆Z
of all measured concentration profiles was made. The
value of ∆Z was estimated using the following
semiempirical formula proposed earlier [24]:
Q
∆Z = 0 . 37 ⋅ d ⋅ ,
S ⋅V
where d is the probe orifice diameter, Q is gas flow rate
through the probe orifice, S denotes probe orifice area,
V is the velocity of incoming gas flow.
The absolute concentrations of the measured species
were determined using the corresponding calibration
coefficients. These coefficients for H2, O2 and H2O
were obtained by direct calibration. For this purpose, the
measurements of the intensities of the base mass peaks
(m/z) 2, 32 and 18 in the gas mixture of known
composition consisting of air and hydrogen and
containing the water vapor heated to the temperature of
200 ○ Fig. 2. Concentration profiles of H and OH radicals. Symbols: measurements; lines: modeling (using kinetic
mechanism [23]) with temperature profiles measured at different distances “probe tip - thermocouple”.
C were performed. Peak intensity of argon
presenting in the air was used as a reference. The
calibration coefficient ki relating to argon was evaluated
using the following formula:
ki = Ci×I40/CAr×Ii, (1)
where Ii and I40 are the intensities of the base peak for ispecies
and for argon, respectively; Сi and СAr are mole
Mole fraction
0,30
0,25
0,20
0,15
0,10
0,05
0,00
T T
H2
H
H2O
O2
0 1 2 3 4
Height above burner, mm
1800
1600
1400
1200
1000
800
600
400
200
0
Temperature, K
Fig. 3. Temperature profile and concentration profiles
of stabile species in calibration flame H2/O2/N2 flame
(φ=1.1, D=0.14). Symbols are experimental data; lines
are modeling using kinetic mechanism [23].
fractions of the species being calibrated and argon,
respectively.
The measurements of H atoms and OH radicals in
the flames were carried out with the energy of ionizing
electrons of 16.2 eV, which allowed the contribution
from other species to the corresponding mass spectra to
be excluded. Calibration of H and OH radicals was
performed using a well known method proposed earlier
[25]. We have already employed it for subatmospheric
hydrogen-oxygen flames [26]. This method is based on
a fact that in the post-flame zone of hydrogen flames
partial equilibrium of three “fast” reactions takes place.
These three reactions are the following:
Н2 + ОН =Н2О + Н (I)
Н2 + О = Н + ОН (II)
О2 + Н = ОН + О (III)
The equilibrium constants K1, K2 and K3 [23]
corresponding to the aforementioned reactions are given
below:
К1=0.113×Т 0.0839 ×е 7680/Т (2)
К2 = 1.8×Т 0.027 ×е -917/Т (3)
К3=302×Т -0.374 ×е -8620/Т (4)
The expressions for H, O and OH in this
approximation are of the following form:
[OH] = K 2 × K 3 × [ O2
] × [ H 2 ] , (5)
[O] = K 1 × K3
× [ H 2]
× [ O2]
, (6)
[ H 2O]
[H] = K1[
H 2 ]
K 2 × K 3 × [ O2
] × [ H 2 ] . (7)
[ H 2O]
In the post flame zone (1.5-4 mm above the burner)
of all hydrogen flames with φ=0.47, 1.1 and 2.0
(D=0.209, 0.09 and 0.077, respectively) studied in this
work, the concentrations of H and OH calculated by
both methods using PREMIX and the detailed kinetic
mechanism [23] and using the above formulas differ.
This is because of relatively low post-flame temperature
(1200-1350 К). For this reason, to carry out the
calibration of radicals we used hotter flame (calibration
flame) with the post-flame temperature of 1600 K (see
Table 1). To determine the calibration coefficients for
H and OH radicals and hence to obtain their absolute
concentration, concentration profiles of H2, O2 and H2O,
as well as temperature profile were measured. They are
shown in Fig. 3. In this figure, the measured profiles are

Mole fraction
0,012
0,010
0,008
0,006
0,004
0,002
0,000
OH
H
0 1 2 3 4 5
Height above burner, mm
Fig. 4. Concentration profiles of radicals in calibration
flame H2/O2/N2 flame (φ=1.1, D=0.14). Symbols are
experimental data; solid lines are modeling with the
detailed chemical kinetic mechanism [23]; dashed lines
are calculation using the approach of partial equilibrium
on three “fast” reactions.
compared with those calculated using kinetic
mechanism [23].
Inserting the values of concentrations of stable
species and equilibrium constants into the formulas (5)
and (7), the concentrations of H and OH in the postflame
zone (at 1.2-1.5 mm) of the calibration flame (φ=
1.1, D = 0.14) above the burner were evaluated. The
concentration profiles of H and OH calculated using the
detailed kinetic mechanism [23] (and measured
temperature profile with the distance “probe tip –
thermocouple junction” of 0.2 mm) and using the
approach of partial equilibrium on three fast reactions
are compared in Fig. 4. A good agreement clearly
observed between them demonstrates a correctness of
selection of the experimental conditions for calibration
of the MBMS-system. Thus, using the data obtained in
the “hot” calibration flame, the calibration coefficients
for H and OH were determined. These coefficients were
used in the following for quantification of absolute
concentrations of H and OH in all flames studied (lean,
near-stoichiometric and rich). Due to a very low
concentration of atomic oxygen in the post-flame zone,
calibration coefficient for it was not determined with
adequate accuracy, so we do not provide in the
following measured concentration profiles of O radicals.
Relative error in determining the concentrations of the
species from the measured mass peak intensities of H
and OH radicals was within the range 40 - 60 %.
Modeling
Flame structure and burning velocity were simulated
using PREMIX and CHEMKIN-II codes [19-21] taking
into consideration thermodynamic and transport
properties of all species, as well as a detailed chemical
kinetic mechanism.
Three different reaction mechanisms for hydrogen
oxidation proposed earlier by Li et al. [1], O’Conaire et
al. [2] and Konnov [23] were used for the modeling.
4
0.18
0.16
H2O 1600
1400
0.14
0.12
T
1200
0.10
1000
0.08
0.06
0.04
H2 O2 800
600
0.02
400
0.00
200
0 1 2 3 4 5
Height above burner, mm
Fig. 5. Temperature profiles and concentration
profiles of stable species in near-stoichiometric
H2/O2/N2 flame (φ=1.1 D=0.09). Symbols are
experimental data; lines are modeling using the
kinetic mechanism [23].
Mole fraction
Mole fraction
Mole fraction
0.25
0.20
0.15
0.10
0.05
0.00
H2
T
O2
H2O
0 1 2 3 4 5
Height above burner, mm
0.25
1400
T
0.20 H2 1200
0.15
1000
800
0.10
0.05
H2O
O2 600
400
0.00
200
0 1 2 3 4 5
Height above burner, mm
1600
1400
1200
1000
800
600
400
200
Fig. 7. Temperature profiles and concentration
profiles of stable species in rich H2/O2/N2 flame
(φ=2.0, D=0.077). Symbols are experimental data;
lines are modeling using the kinetic mechanism [23].
Multi-component diffusion and thermal diffusion
options were taken into account. Adaptive mesh
parameters were GRAD = 0.1 and CURV = 0.5.
Results and Discussion
Figures 5, 6 and 7 show measured concentration
profiles of stable species (H2O, H2, O2) in the nearstoichiometric,
lean and rich flames, respectively. The
profiles simulated using the kinetic mechanism [23] are
also given in these figures. The species concentration
and temperature profiles predicted by the mechanisms
Temperature, K
Temperature, K
Temperature, K
Fig. 6. Temperature profiles and concentration
profiles of stable species in lean H2/O2/N2 flame
(φ=0.47, D=0.209). Symbols are experimental data;
lines are modeling using the kinetic mechanism [23].

of Li et al. [1] and of O’Conaire et al. [2] are very close
to those predicted by Konnov’s mechanism [23]. They
are not shown in Figs. 5-7 to not overcharge them. A
comparison of the measured concentration profiles of
stable species with the simulated ones (Figs. 5-7) shows
OH mole fraction
0.0035
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
experiment
mech. of Li et al. [1]
mech. of O'Conaire et al. [2]
mech. of Konnov [23]
0.0000
0 1 2 3 4 5
Height above burner, mm
Fig. 8. OH concentration profiles in lean H2/O2/N2
flame (φ=0.47, D=0.209). Symbols are
experimental data; lines are modeling.
0.008
Mole fraction
0.006
0.004
0.002
OH
H
mech. of Li et al. [1]
mech. of O'Conaire et al. [2]
mech. of Konnov [23]
0.000
0 1 2 3 4 5
Height above burner, mm
Fig. 9. H and OH concentration profiles in nearstoichiometric
H2/O2/N2 flame (φ=1.1, D=0.09).
Symbols are experimental data; lines are
modeling.
H mole fraction
0.008
0.006
0.004
0.002
experiment
mech. of Li et al. [1]
mech. of O'Conaire et al. [2]
mech. of Konnov [23]
0.000
0 1 2 3 4 5
Distance above burner, mm
Fig. 10. H concentration profiles in rich H2/O2/N2
flame (φ=2.0, D=0.077). Symbols are experimental
data; lines are modeling.
5
that the models used provide a good fit of the
experimental data. Some small discrepancies observed
between the measured and simulated concentrations of
H2O near the burner surface are associated with
variation of the H2O calibration coefficient depending
on flame temperature. This is due to formation of H2O
clusters in molecular beam while sampling from the
low-temperature zone of the flame. Therefore, an
excessive concentration of H2O measured near the
burner surface is related to the experimental errors.
The concentration profiles of H and OH radicals
measured in the lean, near-stoichiometric and rich
flames are given in Figs. 8, 9 and 10, respectively. The
results of numerical calculations using three kinetic
mechanisms [1, 2, 23] are also plotted in Figs. 8-10. A
satisfactory agreement is observed between the
measured and calculated OH concentration profiles in
the lean and near-stoichiometric flames (Figs. 8 and 9).
It should be noted that maximums of the experimental
OH profiles are ~ 0.1 mm shifted downstream relative
to the calculated profiles. Maximum measured
concentration of OH is 2.0×10 -3 and 9.1×10 -4 in the lean
and near-stoichiometric flame, respectively. The models
over-predict them by about 20 - 30 percents. Measured
concentration profiles of H radicals in the nearstoichiometric
and rich flames (Figs. 9, 10) are also in a
satisfactory agreement with the simulated ones. As it
was observed for OH profiles, the measured H
concentration profiles are also shifted downstream by
~ 0.1 mm relative to the calculated ones. Such a shift
can be associated with the errors of taking into account
the probe induced perturbations. Maximum measured
concentration of H is 5.4×10 -3 and 5.9×10 -4 in the nearstoichiometric
and lean flame, respectively. These
values are 16÷30 percents lower than those obtained in
the simulation. Therefore, within the experimental
errors, the experimental data obtained in this work are in
a fairly good agreement with the calculated results. This
indicates the robustness of these models in predicting
the structure of hydrogen-oxygen flames at atmospheric
pressure.
Conclusions
The structure of laminar premixed H2/O2/N2 flames
stabilized on a flat flame burner with different
equivalence ratios at atmospheric pressure was studied.
Probing molecular beam mass spectrometry with soft
electron-impact ionization was used to measure species
concentration profiles in the flames. In this work, for the
first time, concentration profiles of H and OH radicals
in the lean (φ=0.47), near-stoichiometric (φ=1.1) and
rich (φ=2.0) atmospheric flames were measured.
Numerical simulation of the structure of the flames
using three detailed kinetic mechanisms for hydrogen
combustion proposed earlier was carried out in order to
validate them by comparison of the calculated results
with our experimental data. The comparison showed
that these models reproduce quite well the measured
concentration profiles of the species in the flames
studied. This demonstrates a good performance of the

models in predicting the structure of atmospheric
hydrogen flames.
Acknowledgements
The Russian Foundation of Basic Research is
acknowledged for the support of this work by the grant
№ 07-03-01000.
References
[1] Li J., Zhao Z., Kazakov A., Dryer F.L., Int. J.
Chem. Kinet.36 (2004) 566.
[2] O’Conaire M. , Curran H.J., Simmie J.M., Pitz
W.J., Westbrook C.K., Int. J. Chem. Kinet. 36
(2004) 603.
[3] Konnov A.A., Khimicheskaya Fizika 23(8)
(2004)5.
[4] Zsely I.Gy., Zador J., Turanyi T., Proc. Combust.
Inst. 30 (2005) 1273.
[5] Davis A.G., Joshi A.V., Wang H., Egolfopoulos F.,
Proc. Combust. Inst. 30 (2005) 1283.
[6] Del Alamo G., Williams F.A., Sanchez A.L.
Combust. Sci. Technol. 176 (2004) 1599.
[7] Saxena P., Williams F.A., Combust. Flame. 145
(2006) 316.
[8] Eltenton G.C., J. Chem. Phys. 9 (1942) 403.
[9] Foner S., Hudson R., J. Chem. Phys. 21 (1954)
1317.
[10] Bascombe K.N., Proc. Combust. Inst. 10 (1965)
55.
[11] Brown N. J., Eberius K. H., Fristrom R. M.,
Hoyermann K.H. and Wagner H.Gg., Combust.
Flame 33 (1978) 151.
[12] Vandooren J., Bian J., Proc. Combust. Inst. 23
(1990) 839.
[13] O. P. Korobeinichev, V. M. Shvartsberg, S. B.
Il’in, A. A. Chernov and T. A. Bol’shova,
Combustion, Explosion, and Shock Waves 35 (3)
(1999) 239.
[14] Dixon-Lewis G., Sutton M.M., Williams A., Proc.
Combust. Inst. 10 (1965) 495.
[15] Dixon-Lewis G., Sutton M.M., Williams A.,
Proc.Roy.Soc. London. A 317 (1970) 227.
[16] A. A. Paletskii, L. V. Kuibida, T. A. Bolshova, O.
P. Korobeinichev and R. M. Fristrom, Combustion,
Explosion, and Shock Waves 32 (3) (1996) 245.
[17] De Goey, L.P.H., Van Maaren, A., Quax, R.M.
Combust. Sci. Technol. 1992 (1993) 201.
[18] Van Maaren, A., Thung, D.S., de Goey, L.P.H.,
Combust. Sci. Tech. 96 (1994) 327.
[19] Kee R.J., Rupley F.M., and Miller J.A., Sandia
National Laboratories Report, SAND89-8009.
[20] Kee R.J., Grcar J.F., Smooke M.D., and Miller
J.A., Sandia National Laboratories Report,
SAND85-8240.
[21] Lutz A.E., Kee R.J., and Miller J.A., Sandia
National Laboratories Report, SAND87-8248
(1990).
[22] Korobeinichev O.P., Ilyin S.B., Shvartsberg V.M.,
Chernov A.A., Combust. Flame. 118(4) (1999)718.
6
[23] Konnov A.A., Detailed reaction mechanism for
small hydrocarbons combustion. Release 0.5
http://homepages.vub.ac.be/~akonnov/, 2000.
[24] Korobeinichev O.P., Tereshchenko A.G.,
Emelyanov I.D., Rudnitskii A.L., Fedorov S.Yu.,
Kuibida L.V., Lotov V.V., Combustion, Explosion,
and Shock Waves 21(5) (1985) 22.
[25] Warnatz J., Combust. Sci. and Tech. 26 (1981)
203.
[26] Korobeinichev O.P., Shvartsberg V.M.,
Chernov A.A., Combust. Flame 118(4) (1999) 727.
[27] Konnov A.A. Combust. Flame 152(4) (2008) 507.